Gait (source code)
= Gait
A gait is a pattern of coordinated limb motion and contact used in locomotion. In a periodic gait, the order of foot contacts, their relative timing and the duration of each contact characterize a cycle. A <quadrupedal robot> can change gait without changing the number of legs or their basic mechanical arrangement.
For a repeated foot motion with period $T$, the <stance phase> is the interval during which that foot contacts the ground; the <swing phase> moves it between contacts. If stance lasts $T_s$, its contact <duty cycle> is $\beta=T_s/T$. Relative phase specifies how one leg's cycle is shifted from another's. Equal duty cycles do not determine a unique gait because those phase shifts can differ.
For an original four-leg schedule, take front-left, front-right, hind-left and hind-right phases $(0,1/2,3/4,1/4)$ with $\beta=3/4$. Apart from transition instants, three feet contact the floor. A second constructed schedule takes phases $(0,1/2,1/2,0)$ and $\beta=1/2$: diagonal pairs contact together and alternate every half-cycle, illustrating a <trot>. The first schedule has total contact-time sum $4\beta T=3T$ per cycle; the second has $2T$. Summing leg contact times does not mean a stride itself lasts three or two periods.
\Image[/quadruped-gait-contact-schedules.png]
{title=Constructed <quadrupedal robot> gait contact schedules}
{description=Original periodic timing examples. Colored intervals denote <stance phase>, pale intervals <swing phase>. LF and RF are front legs; LH and RH are hind legs. The schedules prescribe contacts, not measured trajectories, ground forces or a proof of stability.}
A real dynamic apparatus illustrates why timing must connect to <forces>. The 2015 MIT Cheetah demonstration adjusted approach strides and motor forces to execute a running jump; its account describes bounding as successive small jumps.<MIT cheetah robot lands the running jump by Jennifer Chu>[See, run, jump: stride adjustment, force planning and bounding description]{ref} The numerical schedules above are not measurements from that experiment.
A <support polygon> is useful for a stipulated static contact model, but a contact schedule alone does not establish static or dynamic feasibility. <Friction>, joint motion, actuator capability and <angular momentum> can matter. Measuring foot-contact times alongside body motion and ground reactions tests whether a planned gait is actually realized. Nonperiodic transitions, starts and obstacle maneuvers additionally require more than one fixed repeating cycle.
Back to article page
